Exploring ElderNet Datasets

Client Info

main takeaways:

• There are 284 Client IDs are duplicated in the Client_info dataset. Example: anon_ID = 625 has 3 values for age_group, T, D and missing
• There is lots of missing demographic data for poverty status, minority status and age group

client_info %>% 
  add_table()

duplicated <- 
  client_info %>%
  group_by(anon_ID) %>%
  tally(sort = TRUE) %>%
  filter(n > 1) 

duplicated %>% 
  add_table()

client_info %>%
  group_by(poverty, minority, age_group) %>% 
  tally() %>%
  mutate(label = paste('poverty:', poverty),
         age_group = ifelse(is.na(age_group), 'missing', age_group),
         minority = ifelse(is.na(minority), 'missing', minority)) %>%
  ggplot(., aes(x = age_group, y = n, fill = minority)) + 
  geom_col() + 
  facet_wrap(vars(label)) + 
  bg_theme() + 
  scale_fill_npg() + 
  labs(title =" ElderNet Clients by Age Group, Poverty Level and Mintory Status")

Care Management

care_mgmt %>% add_table()

Care management encounters per client is right-skewed. The max is 371 encounters, min is 1 and the median number of care management encounters per client is 6.

care_mgmt %>%
  group_by(anon_ID) %>%
  tally(sort = TRUE)  %>%
  add_table()

quantiles <- 
  quantile(care_mgmt %>% group_by(anon_ID) %>% tally() %>% pull(n), probs = seq(0, 1, .10))

quantiles

##    0%   10%   20%   30%   40%   50%   60%   70%   80%   90%  100% 
##   1.0   1.0   1.8   2.0   4.0   6.0  10.0  15.0  26.2  77.9 371.0

care_mgmt %>%
  group_by(anon_ID) %>%
  tally() %>% 
  ggplot(aes(x = n)) +
  geom_histogram(fill = bg_green, alpha = .75, bins = 50)+ 
  bg_theme() + 
  labs(title = "distribution of care management encounters") 

care_mgmt %>%
  mutate(month = floor_date(assistance_date, 'month')) %>%
  group_by(anon_ID, month) %>%
  tally() %>% 
  ggplot(aes(x = month, y = n, group = month)) +
  geom_jitter(color = 'gray', alpha = .55) + 
  geom_boxplot(fill = bg_green, alpha = .75, outlier.alpha = 0) + 
  labs(title = "distribution of care management encounters per month") + 
  bg_theme() + 
  scale_fill_npg() 

There is a decrease present in both Community and Home Visit encounters starting in March 2020. It also appears remote contacts with clients was shifted to more remote contacts without clients once the pandemic hit.

care_mgmt %>%
  mutate(month = floor_date(assistance_date, 'month')) %>%
  group_by(assistance_category, month) %>%
  tally() %>% 
  padr::pad(group = "assistance_category") %>%
  padr::fill_by_value(n, 0) %>%
  ggplot(aes(x = month, y = n, color = assistance_category)) +
  geom_point() + 
  geom_line()+ 
  bg_theme() + 
  theme(legend.position = 'bottom') + 
  labs(color = NULL) +
  scale_color_npg()  + 
  labs(title = "monthly total care management encounters by type of service provided")

The average time spent per encounter varies from month to month.

care_mgmt %>%
  mutate(month = floor_date(assistance_date, 'month')) %>%
  group_by(month, assistance_category) %>%
  summarise(mean_time_spent = mean(amount)) %>%
  padr::pad(group = "assistance_category") %>%
  padr::fill_by_value(mean_time_spent, 0) %>%
  ggplot(aes(x = month, y = mean_time_spent, color = assistance_category)) +
  geom_point(show.legend = FALSE) + 
  geom_line(show.legend = FALSE) + 
  facet_wrap(vars(assistance_category), scales = 'free') + 
  bg_theme() + 
  scale_color_npg()  + 
  labs(title = "mean time spent per encounter type per month")

Pantry

pantry %>% add_table()

There is some repeating cycles present, likely due to holiday baskes (see next chart).

pantry %>%
  mutate(assistance_date = mdy_hm(assistance_date),
         month = floor_date(assistance_date, 'month')) %>%
  group_by(month) %>%
  summarise(total = sum(amount)) %>%
  padr::pad() %>%
  padr::fill_by_value(n, 0) %>%
  ggplot(aes(x = month, y = total)) +
  geom_line(color = bg_green) + 
  geom_point(color = bg_green) + 
  bg_theme() +
  scale_color_npg() +
  labs(title = "total monthly pantry poundage")

pantry %>%
  mutate(assistance_date = mdy_hm(assistance_date),
         month = floor_date(assistance_date, 'month')) %>%
  group_by(month, assistance_category) %>%
  tally() %>% 
  padr::pad(group = "assistance_category") %>%
  padr::fill_by_value(n, 0) %>%
  ggplot(aes(x = month, y = n, color = assistance_category)) +
  geom_line() + 
  geom_point()+ 
  bg_theme() +
  scale_color_npg() +
  labs(title = "Pantry poundage by category")

Volunteers

volunteer %>% add_table()

volunteer %>%
  mutate(appt_date = mdy(appt_date),
         month = floor_date(appt_date, 'month')) %>%
  group_by(month, category) %>% 
  summarise(mean_duration = mean(appt_duration)) %>%
  padr::pad(group = "category") %>%
  padr::fill_by_value(mean_duration, 0) %>%
  ggplot(., aes(x = month, y = mean_duration, color = category)) +
  geom_line(show.legend = FALSE) + 
  geom_point(show.legend = FALSE) + 
  facet_wrap(vars(category), scales = 'free') +
  labs(title = "monthly volunteer duration by category")+ 
  bg_theme() +
  scale_color_npg() 

Ideas

client story: * filterable UI w/ summary of: * number of recent interactions * time spent * pantry items * services utilized * all absolute and relative to other active clients * can track over time…

utilization metric: * time in month + food + care coordination… * track over time to see if more care is needed if metric starts to decline?

Frame ldernet benefits in terms of the alternative

• ElderNet is the dignified alternative - providing access to care and food keeps people in their homes longer
• ElderNet is the preferred alternative - providing access to care and food allows clients to remain in the home longer
• ElderNet is the less expensive alternative - providing access to care keeps up clients health and can help prevent costly care due to delays in care
• ElderNet is the alternative

client <- 
  "210"

care_mgmt_story <- 
  care_mgmt %>%
  # filter(anon_ID == client) %>%
  mutate(month = as.Date(floor_date(assistance_date, 'month'))) %>%
  group_by(anon_ID, month) %>%
  summarise(total_care_mgmt_encounters = n(),
            total_care_mgmt_mins = sum(amount, na.rm = T),
            mean_care_mgmt_mins = mean(amount, na.rm = T))

pantry_story <- 
  pantry %>%
  # filter(anon_ID == client) %>%
  mutate(month = as.Date(floor_date(mdy_hm(assistance_date), 'month'))) %>%
  group_by(anon_ID, month) %>%
  summarise(total_pantry_encounters = n(),
            total_pantry_pounds = sum(amount, na.rm = T),
            mean_pantry_pounds = mean(amount, na.rm = T))

volunteer_story <- 
  volunteer %>%
  # filter(anon_ID == client) %>%
  mutate(month = as.Date(floor_date(mdy(appt_date), 'month'))) %>%
  group_by(anon_ID, month) %>%
  summarise(total_volunteer_encounters = n(),
            total_volunteer_mins = sum(appt_duration, na.rm = T),
            mean_volunteer_mins = mean(appt_duration, na.rm = T))

get_utilziation <- 
  function(client){
    
    client_info %>%
      filter(anon_ID == {{client}}) %>%
      left_join(., care_mgmt_story) %>%
      left_join(., pantry_story) %>%
      left_join(., volunteer_story) %>% 
      group_by(month) %>%
      summarise(encounters = sum(total_care_mgmt_encounters, total_pantry_encounters, total_volunteer_encounters, na.rm = T),
                mins = sum(total_care_mgmt_mins, mean_volunteer_mins, na.rm = T),
                food = total_pantry_pounds) %>% 
      mutate(monthly_total = sum(encounters, mins, food, na.rm = T))

  }

# total <- 
#   map_dfr(unique(client_info$anon_ID), ~get_utilziation(client = .x))

client_totals <-
  client_info %>%
  select(anon_ID) %>%
  left_join(., care_mgmt_story, by = c("anon_ID")) %>%
  left_join(., pantry_story, by = c("anon_ID", "month")) %>%
  left_join(., volunteer_story, by = c("anon_ID", "month")) 

client_totals %>%
  select(anon_ID, month, starts_with("mean")) %>%
  pivot_longer(cols = 3:5) %>%
  group_by(month, name) %>%
  summarise(total = sum(value, na.rm = T)) %>%
  ggplot(aes(x = month, y = total, fill = name)) + 
  geom_area()+ 
  bg_theme() +
  scale_fill_npg() 

client_totals %>%
  select(anon_ID, month, ends_with("encounters")) %>%
  select(-starts_with("mean")) %>%
  pivot_longer(cols = 3:5) %>%
  group_by(month, name) %>%
  summarise(total = sum(value, na.rm = T)) %>%
  ggplot(aes(x = month, y = total, fill = name)) +
  geom_area()+ 
  bg_theme() +
  scale_fill_npg() +
  labs(title = "Encounters")

client_totals %>%
  select(anon_ID, month, ends_with("mins")) %>%
  select(-starts_with("mean")) %>%
  pivot_longer(cols = 3:4) %>%
  group_by(month, name) %>%
  summarise(total = sum(value, na.rm = T)) %>%
  ggplot(aes(x = month, y = total, fill = name)) + 
  geom_area()+ 
  bg_theme() +
  scale_fill_npg() +
  labs(title = "Minutes")

client_totals %>%
  select(anon_ID, month, ends_with("pounds")) %>%
  select(-starts_with("mean")) %>%
  group_by(month) %>%
  summarise(total = sum(total_pantry_pounds, na.rm = T)) %>%
  ggplot(aes(x = month, y = total)) + 
  geom_area(fill = bg_green, alpha = .65)+ 
  bg_theme() +
  labs(title = "total pantry pounds")